I've used the best offer feature that Julien introduced many times but one issue is that it doesn't seem to be possible to combine multiple best offers accepted into a single cart transaction.

In other words each item you submit an offer for which is accepted is considered a distinct separate purchase and you don't gain any of the benefits of getting bonus positive feedbacks for combined purchases. So you can easily make loads of purchases and get very few feedbacks.

Actually it's the opposite. A single transaction with many items would count as a single feedback.

When making several best offers, you will get a feedback for each individual title and get your feedback score higher, faster.

Julien

So how come I received 8 positive feedbacks for 8 best offer purchases and my feedback count increased by absolutely 0?
In other words it was 104 total before any of those additional feedbacks were received and is still 104 afterwards!

I've always found the feedback system on this site to be very strange.
According to LDDB.com I've completed 355 transactions (buying or selling), 253 of which resulted in positive feedback but the total count is 104.
Even taking into account a few cancelled transactions and the fact that not everyone leaves feedback it still doesn't explain the massive discrepancy in feedbacks left versus the total reported.

Here you go!

Question for Julien about count of feedbacks on LDDB

Julien

Which confirms why I posted this new thread.

Despite what you mention about 'When making several best offers, you will get a feedback for each individual title and get your feedback score higher, faster.' that does not happen.

"You will get a feedback accounted towards the bonus scheme" then.

But if you keep buying from the same seller, the feedback will only increase with a log_2(n) speed.

I see 56 positive feedbacks with bram.kalkman, then you received:

+1 on the 1st transaction
+1 on the 4th transaction
+1 on the 8th transaction
+1 on the 16th transaction
+1 on the 32th transaction

You are 8 transactions away from the next +1.

This is the current scheme, when designed people had few transactions among each other.
The log_2(n) cooling might be a little too strong. Maybe "+1 every N transactions" would be more fair.

Julien

For the historical background:

eBay is giving you a +1 on each transaction. Feedbacks are not changeable easily.
Yahoo Auctions is giving you maximum a +1 per seller. Feedbacks can be changed later.

I tried something in between.

Julien

The current feedback system seems to reward buyers who are constantly changing to a NEW seller to purchase a NEW disc or set of discs with EVERY single transaction and vice versa if the buyer leaves positive feedback for the new seller. In my opinion that doesn't remotely model the real world.

In the real world if a person is happy with a transaction then it seems perfectly logical they would return to the same seller and buy more. Both the buyer and seller should reap the rewards for good customer service and loyalty. In the current system they are heavily penalised for doing so AND if they also use the best offers facility they get 0 feedback since each one is regarded as a single transaction. Likewise the good seller is equally penalised.

You had previously given some reasons why you implemented the feedback system the way it is currently as:

1) It gives a higher emphasize on keeping consistently good behavior with different people.

2) It lowers the probability of collusion between buyer/seller to artificially increase counter through several small transactions.

3) It lowers the probability of a buyer setup'ing a seller's shop on only carry transaction with him/herself to create inflated feedback counter and then scam people away based on it.

I think in practice none of these actually stand up to close scrutiny.

As I've demonstrated it doesn't give higher emphasis to a buyer who returns to the same seller or vice versa. In fact it does the exact opposite - it actually lowers the emphasis and encourages the buyer to shop elsewhere since their total feedback count will rise far faster by switching to another seller as they will always get at least 1 positive feedback! My feedback confirms this.

Has there been a single case of anyone (even cararte ) attempting to be a buyer of their own discs to increase their feedback count? I seriously doubt anyone would have the time or patience to do this and if they did it would be easy to spot from information like country, address or ipaddress etc. Don't you do checks on all of us before we setup shops anyway?

I also haven't met a single seller yet who tried to bribe me to purchase more from them in order to gain more positive feedbacks. That has to be one of the weakest reasons I can think of. Money talks in my world....not feedbacks!

I don't doubt the current feedback algorithm involved some interesting mathematical programming but the end result is pure fantasy. I think a simpler algorithm would have sufficed in this case and been a much closer approximation to reality.

I'll try to play with a few more schemes and see how it reshuffles the world

Julien

I think the log_2(n) feedback cooling adjustment for repeat orders sounds perfect to me, for what it's worth. But enough about feedback.

The original question was a feature request to be able to make an offer for multiple items, as a batch. That is something that would be particularly good.

I would like to be able to make offers to sellers on a whole batch at once, to be accepted or declined, or counter-offered as a batch.

Making multiple offers on separate items is no solution because just one or two declined offers can often mean that the shipping-cost benefits of large orders are diluted to the point that it's no longer a good deal for the buyer.

Often the attractiveness of a batch therefore depends on getting the whole batch in other words.

It does shaken things quite a bit to use a 1-for-1 transaction/feedback straight calculation!

Current system:



1-for-1:



My fear of feedbacks manipulation I witnessed on eBay, even got a page on Wikipedia.

However, LDDb sellers are preselected -- and, yes, I turn down about 1/3rd of requests due to complete lack of previous selling experience (LD or other goods) on eBay, Amazon, Yahoo Auctions, or elsewhere -- which should limit the risk attached to these practices.

And I know other fellow LDDb users are always noticing suspicious behavior either on the forum or in shop orders.

So, laserdisc_fan, enjoy your belated Xmas gift in the form of a +158 feedback increase

Julien

It honestly felt like it didn't matter how many transactions I conducted my feedback almost never went up until you adjusted the algorithm to something everyone can actually follow. By way of example prior to you changing this I had received at least 9 more positive feedbacks (since my 104 total) and yet the total was still 104. I knew it was miles out and the same is true for anyone else with a large number of transactions - they simply weren't reflected. Thanks for changing this as it is very straightforward and easy to follow now.

The original request about being able to submit multiple purchases with best offers is actually less important now because the seller can simply compute the total and send a single invoice. There are always manual calculations in working out the total shipping anyway so it is more of a nice to have than essential feature now that you have adjusted the feedback algorithm since that actually addresses the only negative side effect of multiple best offers. Good work and thanks!

With the old algorithm the way to inflate your feedback score was simply to buy say 20 (or more) dirt cheap LDs from a new seller, then move on and buy 20 (or more) dirt cheap LDs from another seller etc. Just keep repeating that pattern never returning to the same seller ever again until you have more feedbacks than the best seller on the site! This approach would have earned you feedbacks for every single purchase + the bonus points without spending much on actual laserdiscs. The only expense being you end up with probably 200 copies of Speed, or Terminator! The irony of it is that the complex algorithm created a simple weakness in the system which could easily be exploited to gain feedbacks far faster than any genuine buyers/sellers....I'm actually surprised no one actually used that!

One final comment is my original total feedback score of 104 would have actually been even lower were it not for the fact that most of feedbacks that counted toward that total came from selling rather than buying. I mostly sell music LDs and for some reason that tended to attract new buyers many times whereas my own buying habits generally involved returning to the same 10 or so trusted sellers so I almost never got feedbacks from buying even though I've bought loads of discs.